From simply leaving a check strip to replicated trials, field-level data collection and analysis can be overwhelming.
For this episode of The Agronomists, host Lyndsey Smith is joined by Jenn Sabourin of Manitoba-based Antara Agronomy, and Aaron Breimer of Deveron, based near Chatham, Ontario. The three talk about field trial set up, data collection, and the all-important doing SOMETHING with the data.
CEU Credit Form
- Disclaimer: we might not actually get to any clips tonight, because our guests for this episode did such a great job prepping us!
- Bremer’s favourite piece of data has to do with soil testing
- Sabourin’s favourite place to start is by asking about management practices
- A slide that includes Nicholas Cage? Stay tuned!
- Confirmation bias: when setting up trials, unconsciously looking for the answer that we were hoping for
- Correlation does not equal causation (Ice cream sales vs shark attacks, Number of people who drowned by falling into a pool correlating with films that Nicholas Cage appeared in)
- Setting up a trial in the spring… does seed get jammed in the ground a little too quick? Important to set the expectations and explain why the trial’s being done.
- REPLICATIONS. That is all. But, really replications are incredibly important especially when the end data set can be affected so much by environmental conditions. (Editor’s note, can speak from personal experience in making the mistake of not replicating enough)
- How far apart can those replications be? Depends on the question or the objective. 20 miles could be a huge radius for some questions; weather could affect the trials too much with that big a range
- Do the numbers make your head explode? Leave it to the experts…?
- Randomized check strips.
- Also, replicating trials over a few years, and not just treatments
- What about using one field-length strip of each treatment? Breimer says you can’t run stats on it.
- What if you’re a newbie, starting out, yield is a great piece of data to collect first, but what’s next? The next piece of info that’s most important to Breimer? Past management practices. Can learn so much from the farmer which can sometimes explain a lot. Sabourin says that equipment widths is really handy info for setting up trials. Treatments that are one full pass around, etc.
- What is the minimum farm size to justify the cost of the technology needed to collect data?
- What’s a good approach for outliers? “Scrubbing” data of weird numbers or really obvious data points that are really out there, they affect data and have to be dealt with a certain way
- Data analysis isn’t easy
- Calibrating equipment is really important.
- With a drought year like 2021, or when there’s excess moisture, where do you draw the line of keeping that data, or punting it?